A Study on Hemodynamic and Brain Network Characteristics During Upper Limb Movement in Children with Cerebral Hemiplegia Based on fNIRS.

Background: Hemiplegic cerebral palsy (HCP) is a motor dysfunction disorder resulting from perinatal developmental brain injury, predominantly impairing upper limb function in children. Nonetheless, there has been insufficient research on the brain activation patterns and inter-brain coordination mechanisms of HCP children when performing motor control tasks, especially in contrast to children with typical development(CD). Objective: This cross-sectional study employed functional near-infrared spectroscopy (fNIRS) to systematically compare the cerebral blood flow dynamics and brain network characteristics of HCP children and CD children while performing upper-limb mirror training tasks. Methods: The study ultimately included 14 HCP children and 28 CD children. fNIRS technology was utilized to record changes in oxygenated hemoglobin (HbO) signals in the bilateral prefrontal cortex (LPFC/RPFC) and motor cortex (LMC/RMC) of the subjects while they performed mirror training tasks. Generalized linear model (GLM) analysis was used to compare differences in activation intensity between HCP children and CD children in the prefrontal cortex and motor cortex. Finally, conditional Granger causality (GC) analysis was applied to construct a directed brain network model, enabling directional analysis of causal interactions between different brain regions. Results: Brain activation: HCP children showed weaker LPFC activation than CD children in the NMR task (t = -2.032, p = 0.049); enhanced LMC activation in the NML task (t = 2.202, p = 0.033); and reduced RMC activation in the MR task (t = -2.234, p = 0.031). Intragroup comparisons revealed significant differences in LMC activation between the NMR and NML tasks (M = -1.128 ± 2.764, t = -1.527, p = 0.025) and increased separation in RMC activation between the MR and ML tasks (M = -1.674 ± 2.584, t = -2.425, p = 0.031). Cortical effective connectivity: HCP group RPFC → RMC connectivity was weaker than that in CD children in the NMR/NML tasks (NMR: t = -2.491, p = 0.018; NML: t = -2.386, p = 0.023); RMC → LMC connectivity was weakened in the NMR task (t = -2.395, p = 0.022). Conclusions: This study reveals that children with HCP exhibit distinct abnormal characteristics in both cortical activation patterns and effective brain network connectivity during upper limb mirror training tasks, compared to children with CD. These characteristic alterations may reflect the neural mechanisms underlying motor control deficits in HCP children, involving deficits in prefrontal regulatory function and compensatory reorganization of the motor cortex. The identified fNIRS indicators provide new insights into understanding brain dysfunction in HCP and may offer objective evidence for research into personalized, precision-based neurorehabilitation intervention strategies.